Anthropic updated Claude Security (public beta for Claude Enterprise) from point-in-time scans to continuous monitoring. After a deep baseline scan, it checks new commits and scans changes based on their risk and complexity, with no fixed schedule. Findings are tracked individually, marked fixed when a fix lands (even if it wasn't made in Claude) and reopened if the issue comes back.

• Projects hold up to eight repositories (GitHub, GitHub Enterprise Server, GitLab and others by token), so it can look for bugs that span services. Each project keeps an editable threat model.
• Findings go out through an API, an MCP connector and webhooks, with Full, Triage and View roles.
• Anthropic says thousands of organizations have found more than a million vulnerabilities since the April beta. That's the vendor's own figure. It's offering up to $25,000 a month in credits for two months through Dec 31.

Worth noting: giving a scanner standing read access to many repos, plus an MCP and webhook path for findings, adds a new high-value integration. Scope its tokens like any other CI identity.